Re: Diagnosing an unstable Qube2

[Andy Ruhl <acruhl%gmail.com@localhost>]

On Mon, Oct 29, 2012 at 12:41 AM, Rémi Zara <remi_zara%mac.com@localhost> wrote:
> Hi,
>
> After some years running quite well, my Qube2 (NetBSD 5.1) started to lock up 
> after a few days, with the serial console unresponsive.
> Usage is regular postgresql building and testing, which is quite processor 
> and memory intensive, backup via rsync.
> I changed the power supply, but that did not solve the problem.
>
> So I upgraded to NetBSD 6.0, and the machine is now really unstable:
> lots of core dumps until the machine is unusable, tar telling that headers 
> are invalid, then succeeding on the same file,
>
> then errors like this:
> sendmail: panic: myfree: corrupt or unallocated memory block
> [1]   Done                    /bin/sh /etc/dai... |
>      Done                    tee /var/log/dai... |
>      Abort trap (core dumped) sendmail -t
>
> How do I go about diagnosing what is the problem ?
> Are there memory tests, disk tests that I could run ?
>
> Thanks for your help, I would be sad to have to decommission this little 
> machine :-)
>
> Rémi Zara

This seems very familiar to me, but I searched and couldn't find what
I was looking for. I seem to remember doing checksums on the same file
repeatedly and getting different results.

I know there were some issues related to disconnecting the fan, and
the theory was that the fan added some stability to the power supply.

What you are describing could be hardware instability, but if it's
really related to upgrading to version 6 then maybe not. It is
probably a good idea to take the machine apart and clean it, and
re-seat the riser card and memory if you haven't done it in a while.

I haven't used my Qube 2 machine in a while, but if I get time I will
try it again.

You might want to copy another list, this one has been pretty quiet.
netbsd-users is probably a good idea.

Andy